perspective-dev/perspective
DataBuild interactive data grids, pivot tables, and charts on large or streaming datasets in the browser, Python, or Rust.
What it does
Perspective is a data grid, pivot table, and charting component for large, real-time, and streaming datasets. It targets developers building user-configurable reports, dashboards, notebooks, and embedded analytics applications. The UI runs as a custom element or a React component, so end users can drag fields and reconfigure queries without new code.
The data engine is written in C++ and compiled to WebAssembly, Python, and Rust. Tables update incrementally and views refresh in real time, with reactive joins across tables, a columnar expression language, and read/write/streaming support for Apache Arrow, CSV, and JSON. For datasets beyond memory, the browser build supports 64-bit memory64 and paging columns to OPFS; native builds can use memory-mapped files.
A symmetric client/server architecture lets the same Client API talk to an in-process engine, a Web Worker, or a remote engine over WebSocket. Server handlers exist for Python (aiohttp, Starlette, Tornado), Node.js, and Rust. Virtual servers translate view configurations into native queries for DuckDB, ClickHouse, PostgreSQL, and Polars, so data can stay where it already lives. A Jupyter widget based on anywidget covers notebook analysis.
Quick start
For browser or app development, install the React component or the Web Component viewer from npm. Both are published under the @perspective-dev scope, with API documentation at React component and Web Component viewer.
For Python, install the package from PyPI, which also provides the Jupyter widget perspective.widget. The Python client documentation is at Python API, and Rust users can follow the perspective crate docs. Before writing code, the browser-based examples on the project site let you load sample datasets and try the grid, pivot, and chart configuration without a local install.
The README does not give a single copy-paste snippet that loads a dataset and renders a view end to end. The authoritative starting points are the user guide and the per-language API references; use the guide to construct your first table and view, then bind it to the React, Web Component, or notebook widget.
Usage constraints
Browser datasets above 4 GB depend on the 64-bit WebAssembly build and OPFS column paging, so availability depends on browser support for those features. Remote and virtual-server setups require a running engine or a connected DuckDB, ClickHouse, PostgreSQL, or Polars service. The project is licensed under Apache-2.0 and is a member of the OpenJS Foundation, which matters if you redistribute the branding or participate in the community.
HotRepo's take
On 2026-11-19, Perspective recorded 100 new stars and rank 11 on the Rust daily list in HotRepo's snapshot. That is a single recorded daily observation, so it shows the project still draws attention in the Rust community but does not establish a longer trend. The value here is the combination of a real streaming query engine and a ready-made analytical UI across WebAssembly, Python, and Rust, which distinguishes it from chart-only libraries.
Written by AI from the project's README, updated . The project's own docs are the source of truth.
Hacker News
You might also like
- Canner/WrenAIOpen-source GenBI engine that gives AI coding agents a governed semantic layer for text-to-SQL and dashboards.
- grafana/grafanaOpen-source platform for querying, visualizing and alerting on metrics, logs and traces from multiple data sources.
- dataease/dataeaseAn open-source BI platform for connecting data sources and building shareable dashboards through drag-and-drop.